Sasha (Augusto) Kielbowicz

Independent Consultant — Agentic AI & Research Software Engineering

I help R&D and Quantitative Finance teams move research-quality models into production without losing mathematical correctness or computational performance. Seven years across JP Morgan, Qontigo, SimCorp, and Mercado Libre. Currently Senior AI Engineer at AstraZeneca (Oncology R&D).

Trust Markers

• Senior AI Engineer at AstraZeneca (Oncology R&D)
• Architected Core Libraries for Qontigo & SimCorp
• 7+ years at tier-1 financial institutions (JP Morgan, Qontigo, SimCorp)
• Published Research in Hydrodynamical Instability Processes

What I Do

Agentic AI Engineering

I build production-grade agent platforms: Model Context Protocol (MCP) integrations, Knowledge Engines with hybrid RAG, automated Evals frameworks, multi-layer agent memory, and LLM security architecture.

Recent work: Designed and shipped the external integration surface and Evals suite for AstraZeneca’s R&D agentic platform, achieving 13%+ reduction in LLM token consumption.

Eliminate Computational Bottlenecks (HPC Sprints)

Your Python or R research prototypes are stalling your pipeline. I optimize numerical bottlenecks using vectorization, caching, and parallelism.

Proven results: 300% performance gains at Axioma/Qontigo, enabling real-time risk calculations in production.

Research Software Engineering (RSE)

Through Phorma — a consultancy I co-founded with Agustín Corbat — we help R&D and Quantitative Finance teams build stable, reproducible, production-grade systems. Scientists keep their research focus; Phorma handles the engineering. Read more about why I created Phorma.

Domain-Specific Language (DSL) Design

I design custom languages that allow Quants to define complex pricing models safely and intuitively, abstracting away the underlying code complexity.

How to Work With Me

I take on a limited number of specialized engagements alongside my primary commitments. Two entry points:

• The Architecture Audit — A focused review of your current pipeline to diagnose computational bottlenecks, LLM integration risks, or RSE gaps. 1–2 hours, actionable findings.
• The Focused Sprint — A 1–4 week intensive targeting a specific bottleneck: numerical optimization, MCP integration, agentic Evals, or DSL prototype.

For team-scale projects or longer R&D engagements, we work through Phorma.

→ Book a call — or reach out on LinkedIn if you’d prefer to introduce yourself first.

Resume / CV

Writing

ak.saxa.xyz/blog — technical posts on agentic AI, RSE, and scientific computing.

Contact

Book a call: calendar.app.google/txzedFUh1p7nvuAX8

LinkedIn: linkedin.com/in/augusto-kiel

GitHub: github.com/akielbowicz